Update README.md

README.md

@@ -33,6 +33,12 @@ You can use the raw model for object detection. See the [model hub](https://hugg
 
 
 ## How to use
+Please note all modes ending with seg or have seg in their name will do segmentation+masks.
+All models having a 'ded' in their name will do a dedection of the image. 
+we are geenrally focussed in providing 8m bassed models . 
+But one can use 8l, 8n or 8x models as these are trained on different image sizes and different image sets.
+Generally the coco and openImage sets from ultralytics are taken for our comfyUI testing. 
+More info and general models for other inferences will be updated  as they get trained.